Children's interpretations of general quantifiers, specific quantifiers and generics

Abstract

Recently, several scholars have hypothesised that generics are a default mode of generalisation, and thus that young children may at first treat quantifiers as if they were generic in meaning. To address this issue, the present experiment provides the first in-depth, controlled examination of the interpretation of generics compared to both general quantifiers (‘all Xs’, ‘some Xs’) and specific quantifiers (‘all of these Xs’, ‘some of these Xs’). We provided children (3 and 5 years) and adults with explicit frequency information regarding properties of novel categories, to chart when ‘some’, ‘all’ and generics are deemed appropriate. The data reveal three main findings. First, even 3-year-olds distinguish generics from quantifiers. Second, when children make errors, they tend to be in the direction of treating quantifiers like generics. Third, children were more accurate when interpreting specific versus general quantifiers. We interpret these data as providing evidence for the position that generics are a default mode of generalisation, especially when reasoning about kinds.

Keywords:

• generics
• children
• quantifiers

Acknowledgements

We are grateful to the parents, teachers, and children who participated in this research, including the staff at Generations Together and Tutor Time preschools. We thank Natalie Delave and Melanie Greenspan for their assistance in collecting and entering the data, and David Barner and an anonymous reviewer for comments on a prior draft.

Notes

1. There are also two notable respects in which children systematically differ from adults, in their interpretation of specific quantifiers. The first concerns the extent to which children are sensitive to scalar implicatures (Grice, 1975) – for example, the extent to which they take ‘some’ to be incompatible with ‘all’. Adults are more likely than children to think that, e.g. ‘some of these coins are in the box’ is false when each and every coin is in the box (Noveck, 2001). Second, young children's interpretations of specific universal quantifiers are susceptible to spreading. That is, if children are shown a picture in which, e.g. five rabbits each have a carrot, but in which there are two additional carrots with no attendant rabbit, they exhibit a tendency to deny that ‘every rabbit has a carrot’ (Drozd, 2005; Roeper, Pearson, & Grace, 2011). (Even though ‘every rabbit has a carrot’ does not contain an explicit restriction to these rabbits, and so may appear syntactically like a general quantifier, the noun phrase ‘every rabbit’ (unlike ‘all rabbits’) is interpreted as applying only to the contextually salient set of rabbits (e.g. Stanley & Szabó, 2000), and therefore functions semantically like a specific quantifier.).

2. One unexpected result is that many children had difficulty with the 0% frequency level, in which cards depicted no instances with the property. When the property was absent, many children reinterpreted other features as if they were the targeted feature (e.g. saying that a yellow colour looked a little bit green, or that an animal with short hair had big hair). Responses were much more consistent when the target feature was present in at least one instance on the card (i.e. all other response levels).

Additional information

Funding

This research was supported by research funds from NICHD Grant [HD-36043] to Gelman, and NSF Grant [grant number BCS-1226942] to Leslie.